Biochimica et Biophysica Acta, 1202 (1993) 129-134 129 © 1993 Elsevier Science Publishers B.V. All rights reserved 0167-4838/93/$06.00

BBAPRO 34561

Change of substrate specificity by chemical modification of lysine residues of porcine pancreatic a-amylase

Hiroshi Yamashita, Hiroshi Nakatani and Ben'ichiro Tonomura Department of Food Science and Technology, Faculty of Agriculture, Kyoto University, Kitashirakawa, Kyoto (Japan)

(Received 25 September 1992) (Revised manuscript received 17 December 1992)

Key words: Alpha amylase; Amylase; Enzyme modification; Substrate specificity; (Pancreas)

Lysine residues of porcine pancreatic a-amylase (PPA) were modified with trinitrobenzenesulfonate (TNBS). 6 out of 21 lysine residues were modified per PPA molecule. Amylase activity (hydrolysis of the a-l,4-D-glucoside bond) was decreased to about 50% of the native enzyme, as judged from the kca t value at pH 6.9 after the modification, whereas maltosidase activity (hydrolysis of p-nitrophenyl-a-o-maltoside producing p-nitrophenol and maltose) was increased to about 250%. The increase in maltosidase activity of the modified PPA was due to the increase in kcat, but not to the decrease in K m. Modification of PPA with five kinds of acid anhydrides also caused the same effect as TNBS, including the number of modified lysine residues. The degree of increase in maltosidase activity was fairly proportional to the volume of the incorporated modification reagent. A modification protection study in the presence of maltotriitol (G3OH), which protected two out of six modifiable lysine residues against modification, suggested that a lysine residue at the substrate-binding site contributes to the change of substrate specificity.

Introduction

Porcine pancreatic a-amylase (PPA) (1,4-a-D-glucan glucanohydrolase, EC 3.2.1.1) catalyzes the hydrolysis of the a-l,4-glucoside bond in starch and maltooligo- saccharides (amylase activity) and hydrolysis of p- nitrophenyl-a-D-maltoside (G2PNP) producing p- nitrophenol and maltose (maltosidase activity). Two aspartic residues (Asp-197 and Asp-300) are proposed as the catalytic residues by an X-ray crystallographic study [1], and contribution of an additional histidine residue to the catalysis was suggested recently by the analysis of pH profiles of catalytic activity [2]. PPA requires CI- as a specific activator, which enhances amylase activity about 30-times at neutral pH [3] and causes shift of the optimum pH of the catalytic activity from 5.2 to 6.9 [4].

The active site of PPA has five subsites for glucose residues of substrates, and the catalytic site is located between the subsites 3 and 4, starting from the site for the non-reducing-end side of the substrate [1,5].

Correspondence to: H. Nakatani, Department of Food Science and Technology, Faculty of Agriculture, Kyoto University, Kitashirakawa, Kyoto 606-01, Japan.

Amino-acid residues around the catalytic site are well- conserved for animal, plant and microbial a-amylases [6,7].

Chemical modification of histidine residues of PPA with diethylpyrocarbonate (DEP) caused decrease of amylase activity [8,9] and increase of maltosidase activ- ity in the presence of CI- [10]. Substrate-selective inhibition by a proteinaceous inhibitor was observed in the histidine-modified PPA [11,12]: the inhibitor from Phaseolus vulgaris strongly inhibited both amylase and maltosidase activities of the native enzyme, but it inhib- ited only the amylase activity of the histidine-modified enzyme. Chemical modification of Taka-amylase A (a mold t~-amylase from Aspergillus oryzae) with DEP also showed similar effects to PPA: decrease of amy- lase activity and increase of maltosidase activity [13]. His-201 in PPA, which is located around subsite 4, was suggested to be chiefly concerned with the change of substrate specificity by the chemical modification [10], and the study on the mutant human pancreatic a- amylase showed that only His-201 at the active site played an important role for this effect [14]. However, to date the molecular mechanism of the effects of the chemical modification has been unclear. Here, we de- signed further chemical modification studies to clarify the molecular mechanism of the enhanced maltosidase activity: the modification of lysine residues of PPA.

130

Based on the chemical modification studies and on the structure of active site of PPA, we conclude here that steric effect and a local hydrophobic environment at the selected part of the active site of PPA intro- duced by the chemical modification are important for the change of substrate specificity.

Materials and Methods

Materials. Isozyme I of porcine pancreatic a-amylase (called PPA throughout this paper) was purified from pancreatin purchased from Sigma [15,16]. The con- centration of the enzyme was determined by using AI~ = 24.0 cm-1 [17] and its molecular mass was 55 280 n m

kDa [18]. G2PNP was purchased from Boehringer- Mannheim and maltotriitol (G3OH) and maltohexaose (G 6) from Hayashibara Biochemical Laboratories. Mal- tohexaitol (G6OH) was prepared by the reduction of G 6 with sodium borohydride [2,19]. Trinitrobenzene- sulfonate (TNBS) and other reagents were purchased from Nacalai Tesque and Wako.

Chemical modification of lysine residues with TNBS. Lysine residues were modified in 10 mM phosphate buffer (pH 8.1) at 25°C with 3 mM (0.1% (w/v)) TNBS solution in water [20]. Final concentration of PPA and TNBS were 10/~M and 1 mM, respectively. After 9 h, excess L-lysine (10 mg/ml reaction mixture) was added to terminate the reaction. Then, the reaction mixture was dialyzed against a chloride-free buffer (20 mM phosphate buffer (pH 6.9)) at 4°C to remove excess lysine, TNBS and their reaction product. The number of modified lysine residues per enzyme molecule was determined spectrophotometrically by the change of absorbance at 340 nm, derived from the N~-tri - nitrophenyl group of the lysine residue. The molar absorption coefficient of the N~-trinitrophenyl group at 340 nm was 14 000 M- 1 cm- 1 at pH 6.9.

C1- concentration of the enzyme, substrate and buffer solutions, when no external CI- was added, were measured with the perchlorate method [21,22] in order to confirm them to be practically chloride-free.

Chemical modification of lysine residues with acid anhydrides. Five kinds of acid anhydrides: acetic, bu- tyric, maleic, 3-nitrophthalic and succinic anhydride, were used as modification reagents. Modification of lysine residues was carried out in 130 mM phosphate buffer (pH 6.7) at 25°C. Acid anhydrides were dis- solved or diluted with dioxane before use [23,24]. Final concentration of PPA and acid anhydrides was 10 ~M and 1-20 mM, respectively. After 1 h of incubation, 50 mM ethylenediamine was added to terminate the reac- tion. Then the reaction mixture was applied to a Sephadex G-25 column (20 mm ID x 20 cm) equili- brated with 20 mM phosphate buffer (pH 6.9) contain- ing 25 mM NaC1 to remove excess reagents. The num- ber of modified lysine residues per enzyme molecule

was determined by measuring the number of unmodi- fied lysine residues in urea-denatured PPA with TNBS. After being dialyzed against 5% NaHCO 3 (pH 8.5), the modified PPA solution (1 ml) was poured into 1 ml of 8 M urea solution and incubated at 40°C for 30 min. Then, 1 ml of 15 mM TNBS solution was added, and incubation was carried on for another 2 h. The reaction mixture was applied to a Sephadex G-25 column (20 mm ID X 20 cm) equilibrated with 1.7% NaHCO 3 con- taining 2.7 M urea to remove excess TNBS. Incorpo- rated TNBS per molecule was measured spectrophoto- metrically as above. For each sample treated in this way, subtracting the number of incorporated TNBS from that of the native PPA gives the number of lysine residues that were modified with acid anhydrides.

Protective modification. In order to protect the ac- tive site from modification, G3OH (0-200 mM) was added in the modification-reaction mixtures with TNBS and 3-nitrophthalic anhydride.

Enzyme assay. PPA-catalyzed reactions were carried out in 20 mM phosphate buffer, containing 0 or 25 mM NaCI (pH 6.9) at 30°C. Amylase activity using soluble starch or G6OH as a substrate was determined from the reductometry of its hydrolysate by the neocuproine method [25]. Cleavage pattern of G6OH by lysine-mod- ified PPA was investigated using Toyo Soda High Speed Liquid Chromatography HLC-802UR. The conditions for G6OH-hydrolysis analyzed by HPLC were as fol- lows: concentrations of G6OH and CI-, 2 mM and 25 mM, respectively; temperature, 30°C; buffer, 20 mM phosphate (pH 6.9); reaction time, 20 min. The condi- tions for separation were as follows: column, Tosoh G-2000PW (7.5 mm ID X 600 mm x 2); column tem- perature, 44°C; eluant, distilled water; flow rate, 1.1 ml/min; sample volume, 100 #1. Maltosidase activity, i.e., hydrolysis of G2PNP, was measured from the increase of absorbance at 400 nm due to p-nitrophenol liberated, which was continuously recorded on a Shi- madzu UV-2200 spectrophotometer. The molar ab- sorption coefficient of p-nitrophenol is 9470 M- ~ cm- 1 at 30°C (pH 6.9) [10]. The pH profile of maltosidase activity for the TNBS-modified PPA was examined by using the buffer system according to Ishikawa et al. [2]. The kinetic parameters (kca t and K m) were deter- mined using the nonlinear least-squares method with Taylor expansion [26].

Results

Chemical modification of lysine residues with TNBS Under the conditions employed, 5.8 (+0.2) out of

21 lysine residues were modified. TNBS can also react with a-amino and thiol groups. However, considering that the a-amino group at the N-terminus of PPA participates in forming pyroglutamate residue and that two free thiols in PPA is not reactive with 5,5'-dithio-

131

TABLE I

Kinetic parameters of native and lysine-modifted PPA in the absence and presence of CI - at 30°C (pH 6.9)

For the determination of amylase activity, soluble starch was used as a substrate. For the determination of maltosidase activity, G2PNP was used as a substrate. Lysine residues were modified with TNBS.

Amylase activity Maltosidase activity

Km × 102 (%) kcat (s -1) K m (mM) kca t × 103

[CI- ] (mM): 0 25 0 25 0 25 0 25

Native 2.04+0.26 1.155-0.11 32.5 5-1.2 8205-19 13.2+0.1 7.055-0.41 9.65+1.12 79.2+0.1 Lys-modifiedenzyme 1.99+0.32 1.275-0.15 9.49+0.51 367+14 11.4+1.2 12.0 +0.6 25.0 5-0.1 202 5-4

bis-(2-nitrobenzoic acid) (DTNB) in the absence of EDTA [27], TNBS would react only with e-amino groups of lysine residues. The kca t value of maltosidase activity was increased up to 250% of that of the native enzyme by the lysine-modification, whereas that of amylase activity was decreased to 45% of the native enzyme (Table I). K m values for both activities were not significantly affected by the modification.

C1- concentrations of 0.25% G2PNP solution, 0.3% soluble starch solution and 20 mM phosphate buffer solution were estimated to be 4.8/zM, 22.1/~M and 0.5 /zM, respectively. These could be regarded practically chloride-free, since a dissociation constant of chloride- bound PPA was reported to be 0.38 mM at 30°C [3]. Addition of CI- to the lysine-modified PPA activated both amylase and maltosidase activities. The degree of activation of each activity by C1- was almost equal to that of the native enzyme (Table I): substrate-selective activation effect, which we previously reported with histidine-modified PPA [10], was not observed. The optimum pH for maltosidase activity of the enzyme was affected neither by the modification of lysine residues nor by the addition of CI- (data not shown).

Chemical modification of lysine residues with acid anhy- drides

About six lysine residues were modified by each acid anhydride, as well as with TNBS. Kinetic parameters of

amylase activity were estimated by using G6OH , which is one of the simplest substrate for amylase activity (Table II). G6OH is uniformly hydrolysed to mal- totriose and G3OH by PPA. Lysine modification by TNBS or each acid anhydride did not change this action pattern. The apparent effects of acid- anhydride-modified PPA on kinetic parameters were the same as the case of TNBS-modified PPA: increase of the kca t value for maltosidase activity and decrease of the value for amylase activity. Modified enzymes treated with hydroxylamine according to Riordan and Vallee [28] showed no significant change in activity compared with those before the treatment. Hence, the effect of side reactions of acid anhydrides with hy- droxyl and thiol groups would be ruled out. There was a fair correlation between the van der Waals volume of the incorporated group by chemical modification and the logarithm of kca t value for maltosidase activity shown in Table II (Fig. 1). Each incorporated volume was estimated, according to Bondi [29]. CI- also acti- vated both activities of each acid anhydride-modified PPA, and the degree of activation was similar to that of TNBS-modified PPA.

Protective modification One lysine residue was protected from TNBS-mod-

ification in the presence of about 10 mM of G3OH , and one more was protected at higher concentrations

TABLE II

Kinetic parameters of lysine-modified PPA in the presence of 25 mM CI - at 30°C (pH 6.9)

The final concentration of each modification reagent was as follows: 1 mM for TNBS; 6 mM for 3-nitrophthalic anhydride; 10 mM for maleic anhydride; 20 mM for succinic anhydride; 2 mM for butyric anhydride; 20 mM for acetic anhydride. Substrates used for amylase and maltosidase activity are GrOH and G2PNP, respectively, n, the number of the modified lysine residues per enzyme molecule.

Modifier n Amylase activity Maltosidase activity

K m X 10 (mM) kca t (s -1) K m (mM) kca t × 103 (s -1)

(Native) 0 3.10 + 0.44 905 + 49 7.05 + 0.41 79.2 5- 0.1 TNBS 5.8+0.2 3.635-0.45 4655-23 10.8 5-0.49 196 5- 4 3-Nitrophthalic anhydride 5.65-0.3 4.085-0.31 5075-16 8.80+1.02 307 5-16 Maleic anhydride 5.4 + 0.3 2.95 5- 0.13 622 5-12 6.59 5-1.47 137 + 12 Succinic anhydride 6.1 5- 0.1 3.56 5- 0.81 462 + 41 6.65 + 2.11 150 + 18 Butyric anhydride 5.35-0.2 2.805-0.31 6105-24 7.515-0.11 196 + 1 Acetic anhydride 6.9 5- 0.2 3.11 + 0.45 301 5- 16 10.2 + 0.40 129 5- 2

132

-0.4

-0.6

~ -0.8

-1.0

-1.2 0

i i i i

• I . . t 1 / ©

2'0 4'o do do loo van der Weals volume of the

Incorporated group (cm a/mol) Fig. 1. Correlation between the van der Waals volume of incorpo- rated modifier and logarithm of kca t value for maltosidase activity. ((3), unmodified; ([]), acetyl group; ( • ) , butyryl group; ( • ) , succinyl group; (zx), maleyt group; ( 0 ) , 3-nitrophthalyl group; (C)), 2,4,6-tri- nitrophenyl group. Dashed line represents linear regression of the

plot. The correlation coefficient is 0.89.

400

300

._~ 200 ¢,) m 100

| | i i i

0' o 5'0 2;o [G~OH] (mM)

16 g ~ s~E 4 e ~ 3 ~ N

~ E 2 ~ N o

0 E

Fig. 2. Protective effect of G3OH on lysine-modification with TNBS. ( • ) , the number of modified lysine residues; (zx), amylase activity ([soluble s tarch]=0.3%); (©), maltosidase activity ([G2PNP] = 0.25%). Relative activity is represented using the initial rate of the reaction per mol of the native enzyme with the substrate concentra-

tion shown above as standard (100%).

(approx. 200 mM; Fig. 2). The first protection caused increase of both amylase and maltosidase activities as compared with the modified enzyme without protec- tion, whereas the second protection caused decrease of maltosidase activity and increase of amylase activity (Fig. 2). These changes of activities were due to changes

TABLE III

Protective effect of G30H on lysine-modification of PPA by 3- nitrophthalic anhydride and TNBS

Modification time with 3-nitrophthalic anhydride was shortened to 5 min in order to maximize the protective effect.

[G3OH] Maltosidase activity a (%)

(mM) b 3-Nitrophthalic anhydride

TNBS

(Nat~e) 100 100 0 240 242

10 239 299 200 93 135

a Relative value of the initial rate against the native enzyme ([G2PNP] = 0.25%).

b This column represents the concentration in the modification-reac- tion mixture.

of kca t values. This '2 lysine-protected and 3.7 lysine- modified' PPA had almost the same activity as the native enzyme. Modification with 3-nitrophthalic anhy- dride was also protected by G3OH in the similar way to TNBS (Table III).

Discussion

Kinetic mechanism of enhanced maltosidase activity The scheme of hydrolysis of G2PNP by PPA is

shown below (Scheme I):

K s

E + S ~ ES

r ~ ES'

k2

Scheme I

, E + P

where E represents PPA (either native or lysine-modi- fled); S, G2PNP; P, products (maltose and p- nitrophenol); ES, productive PPA-G2PNP complex; ES', nonproductive PPA-G2PNP complexes; K s and K~, dissociation constants of productive and nonpro- ductive PPA-G2PNP complexes, respectively and k 2 is the first-order rate constant of product-formation step. The Michaelis parameters, kca t and gm, are obtained as follows:

Ksk2 kca t (1)

K s + K"

KSKs K m (2)

K s + K s

The main binding mode of G2PNP to PPA is a nonproductive mode occupying subsite 1, 2 and 3 (Fig. 3e) as previously reported [30]. Lysine modification would not change this binding mode, judged by the same way as the native enzyme (Yamashita, H., unpub- lished data). Therefore, Eqns. 1 and 2 can be reduced under the assumption of K" << K s to Eqns. 3 and 4, respectively:

g t

kcat = --s k2 (3) Ks

K m = K; (4)

Increase of k 2 o r decrease of K s, or both of them would cause the increase of kca t as has been observed in this case. Increase of k 2 means that the chemical modification stabilizes the transition state to decrease the activation energy for catalysis, whereas decrease of K s means that the affinity of G2PNP to PPA in the productive binding mode is enhanced due to the incor- porated modification reagents. Increase of K' , which means that the nonproductive binding of G2PNP and

133

(b)~~~-

(d) ~ "

Io, I J

Fig. 3. Schematic representation of the binding modes of substrates and ligands to PPA. (a), subsite pattern of PPA. Numbers are assigned from the subsite for non-reducing-end side of substrates to reducing-end side; (b), binding mode of soluble starch. (0), glucose residue; (c), productive binding mode of G6OH. ( c ), sorbitol residue; (d), productive binding mode of G2PNP. (0), p-nitrophenyl group; (e), nonproductive binding mode of G2PNP; (f), binding mode of G3OH at low concentration (approx. 10 raM); (g), binding mode of

G3OH at high concentration (approx. 200 raM).

PPA becomes weaker, also causes increase of kca t. However, this would not be the present case, since g m values of both amylase and maltosidase activities of PPA, which would reflect the main binding mode of the substrate to the enzyme (Eqn. 4), were not signifi- cantly changed by the lysine modification (Table I). At this stage, it is difficult to fix the mechanism.

Roles of modified lysine residues in the change of sub- strate specificity

In the productive binding mode for soluble starch or G6OH, all five subsites are occupied with glucose residues (Fig. 3b,c), whereas glucose residues occupy subsites 2 and 3 and p-nitrophenyl group occupies subsite 4 in the productive binding mode of GEPNP (Fig. 3d). This difference must be reflected in the change of substrate specificity induced by the chemical modification carried out in this study. The logarithm of the enhancement of maltosidase activity was fairly pro- portional to the van der Waals volume of groups incor- porated to lysine residues (Fig. 1). This suggests that the incorporated volume would decrease linearly the Gibbs free energy of activation in the catalytic process. (dAG*/dAV) is estimated as 0.027 k J / c m 3 from the slope of the line on Fig. 1. This result also suggests at least two possible roles of the modified lysine residue: the modified lysine residue assists to form a suitable pocket for the p-nitrophenyl moiety of G2PNP to be bound to subsite 4, or it interacts more tightly with the p-nitrophenyl moiety of GEPNP due to its enhanced hydrophobicity, which would be fairly proportional to the size of the incorporated group as shown in case of

side chains of amino acids [31]. Another probable mechanism is that the modified lysine residues might affect the catalytic residues indirectly.

On the other hand, there was no correlation be- tween volume of incorporated groups and effect on amylase activity: the kea t value for each modified en- zyme was 30-60% of the native enzyme without signifi- cant change of K m value. This means that lysine modification would destabilize the transition state, sug- gesting that the lysine residue in the active site might interact with the transition state by making hydrogen bonds. Modification of a lysine residue would weaken its capacity of making hydrogen bonds, due to loss of a positive charge at its E-amino group. This effect would be independent of the group incorporated. Hence, amylase activities of lysine-modified PPA are almost similar.

Although roles of other modifiable lysine residues are not clear, it is evident that they may partially disturb both amylase and maltosidase activities, and that they are not important for the increase of maltosi- dase activity.

The effect of lysine modification on the change of substrate specificity of PPA was independent of CI- (Table I), in contrast to the histidine-modified PPA which changed substrate specificity only in the pres- ence of CI- [10]. Thus, it is shown that CI- contributes sensitively to catalytic activities depending on the na- ture of modified residues. This would reflect the molecular mechanism of activation of PPA by CI-.

Position of modified lysine residues which affect the enzyme activity

Chemical modifications with TNBS and acid anhy- drides caused decrease of amylase activity and increase of maltosidase activity of PPA. When G2PNP is pro- ductively bound to PPA, the p-nitrophenyl moiety of the substrate occupies subsite 4 of the enzyme (Fig. 3d). On the other hand, a glucose residue occupies subsite 4 when starch or maltooligosaccharides are in their productive binding modes to PPA (Fig. 3b,c).

As shown in Fig. 3f and g, G3OH, which was used to protect the active site of PPA, binds to subsite 1, 2 and 3 at low concentration with an inhibition constant of 0.3 mM [30], and one more molecule binds to subsite 4 and 5 at higher concentration [32]. From the results of protective modification, G3OH could not protect the lysine residues essential for the change of substrate specificity from the modification with TNBS until the concentration became higher than 10 mM (Fig. 2). This effect was also observed in the case of 3-nitrophthalic anhydride modification (Table Iii). These results sug- gest that the lysine residues essential for the change of substrate specificity should be located around subsite 4 and 5. Chemical modification of histidine residues of PPA and site-directed mutagenesis of histidine residues

134

of h u m a n pancrea t ic a-amylase also showed increase of maltosidase activity [11,12,14]. The residue to en- hance maltosidase activity by modif icat ion was assigned to be His-201 which is located a round subsites 4 and 5, sequential ly next to Lys-200. This Lys-His sequence would be re la ted to the similar effects be tween the lysine-modified and the his t id ine-modif ied P P A on the change of substrate specificity. Therefore , at present ,

Lys-200 is the most probable res idue cont r ibu t ing to the change of substrate specificity by chemical modifi- cation. The role of Lys-200 of h u m a n pancrea t ic a- amylase is now u n d e r invest igat ion using the m u t a n t

enzymes.

References

1 Buisson, G., Duee, E., Haser, R. and Payan, F. (1987) EMBO J. 6, 3909-3916.

2 Ishikawa, K., Matsui, I., Honda, K. and Nakatani, H. (1990) Biochemistry 29, 7119-7123.

3 Levitzki, A. and Steer, M.L. (1974) Eur. J. Biochem. 41,171-180. 4 Wakim, J., Robinson, M. and Thoma, J.A. (1969) Carbohydr.

Res. 10, 487-503. 5 Robyt, J.F. and French, D. (1970) J. Biol. Chem. 245, 3917-3927. 6 Friedberg, F. (1983) FEBS Len. 152, 139-140. 7 Nakajima, R., Imanaka, T. and Aiba, S. (1986) Appl. Microbiol.

Biotech. 23, 355-360. 8 Hoschke, A., Laszlo, E. and Hollo, J. (1980) Carbohydr. Res. 81,

145-156. 9 Elodi, P. (1972) Acta Biochim. Biophys. Acad. Sci. Hung. 7,

241-245. 10 Yamashita, H., Nakatani, H. and Tonomura, B. (1991) J. Biochem.

110, 605-607. 11 Nakatani, H. (1988)Arch. Biochem. Biophys. 263, 364-368.

12 Ishikawa, K. and Hirata, H. (1989) Arch. Biochem. Biophys. 272, 356-363.

13 Kita, Y., Sakaguchi, S. Nitta, Y. and Watanabe, T. (1982) J. Biochem. 92, 1499-1504.

14 Ishikawa, K., Matsui, I., Honda, K. and Nakatani, H. (1992) Biochem. Biophys. Res. Commun. 183, 286-291.

15 Loyter, A. and Schramm, M. (1962) Biochim. Biophys. Acta 65, 200-206.

16 Marchis-Mouren, G. and Pasero, L. (1967) Biochim. Biophys. Acta 140, 366-368.

17 Elodi, P., Mora, S. and Krysteva, M. (1972) Eur. J. Biochem. 24, 577-582.

18 Sakano, Y., Takahashi, S. and Kobayashi, T. (1983) J. Jpn. S0c. Starch Sci. 30, 30-37.

19 Wolfrom, M.L. and Thompson, A. (1968) Methods Carbohydr. Chem. 2, 65-68.

20 Haynes, R., Osuga, D.T. and Feeney, R.E. (1967) Biochemistry 6, 541-547.

21 Utsumi, S. (1952) J. Chem. Soc. Jpn. 73, 835-841. 22 Tomonari, A. (1962) J. Chem. Soc. Jpn. 83, 693-695. 23 Butler, P.J.G., Harris, J.I., Hartley, B.S. and Leberman, R. (1969)

Biochem. J. 112, 679-688. 24 Montelaro, R.C. and Rueckert, R.R. (1975) J. Biol. Chem. 250,

1413-1421. 25 Dygert, S., Li, L.H., Florida, D. and Thoma, J.A. (1965) Anal.

Biochem. 13, 367-374. 26 Sakoda, M. and Hiromi, K. (1976) J. Biochem. 80, 547-555. 27 Whitaker, J.R., Finardifilho, F. and Lajolo, F.M. (1988) Biochimie

70, 1153-1161. 28 Riordan, J.F. and Vallee, B.L. (1967) Methods Enzymol. 11,

570-576. 29 Bondi, A. (1968) Physical Properties of Molecular Crystals, Liq-

uids and Glasses, Wiley, New York. 30 Yamashita, H., Nakatani, H. and Tonomura, B. (1992) J. Biochem.

111, 182-185. 31 Chothia, C. (1974) Nature 248, 338-339. 32 Kondo, H., Nakatani, H. and Hiromi, K. (1990) Carbohydr. Res.

206, 161-166.